A Russian AI system can read minds and decode human brain waves
- Tram Ho
It uses only EEG, non-invasive, cheap and easy to apply.
In a new study published in bioRxiv, Russian scientists say they have created an artificial intelligence system that can read and decode human brain waves.
Specifically, it consists of an EEG, which captures nerve signals emanating from the brain, from which the artificial intelligence can draw quite accurately the images that people wear. are seeing in real time.
The best thing about this mind-reading technology is that it is non-invasive, cheap and easy to use. Any good person can use it.
Often, in order to develop brain-computer interfaces that allow reading of minds, many scientists will choose to use a functional magnetic resonance resonator (fMRI) or surgery to plug the electrodes directly. into the patient’s brain.
However, a team of researchers between the Moscow Institute of Physics and Technology and Russia’s Neurobotics Group argues that both technologies have major shortcomings. While electrode implantation is a powerful invasive method that can only be performed in patients, fMRI is limited by the cost of the device being too high, too bulky and difficult to apply into practice.
To solve the problem, Grigory Rashkov and his team at the Moscow Institute of Physics and Technology chose EEG, a basic brainwave recognition technique to use for the brain interface – my computer.
EEG requires only paste or apply electrodes to the scalp, not invasive but still very effective in receiving nerve signals emitted from the brain.
In his research, Rashkov recruited a number of volunteers and included them in the three stages of the experiment.
In the first stage, volunteers only needed to sit in front of a 20-minute computer screen to watch the continuous 10-second videos.
These videos are categorized into 5 topics: abstract shapes, waterfalls, human faces, movement mechanisms and sport bikes. By recording the EEG signal, the scientists can know with each picture, how the brainwave pattern was emitted.
Next to stage two, scientists have developed two artificial intelligence systems to find ways to recreate brain waves recorded from EEG back into images. They must calculate to decode the signals, and at the same time eliminate noise.
Phase three is the testing phase. Volunteers are back on EEG devices but this time, they will be shown completely new videos in each topic. The decoding system has actually reproduced the images that volunteers see from their brainwaves, with the matching of each category sometimes reaching up to 90%.
How can human thought be recreated?
” Many scientists once thought that studying brain processes through EEG was like finding the internal structure of a steam engine by analyzing the smoke track emitted by the locomotive.” , Grigory Rashkov said.
“At first, we didn’t expect it [the EEG signal] would contain enough information to reproduce a portion of the image someone was observing. However, it turned out to be entirely possible. ”
Rashkov further emphasized that his system can help create affordable brain-computer interfaces, and in particular, without surgery. For example, American tech billionaire Elon Musk mentioned a similar interface, but it requires surgery to plug small electrodes into the user’s brain.
Such a technology faces complex challenges, such as the dangers of surgery and the oxidation risks of electrodes that often destroy them in just a few months, Rashkov said.
Now, his new technology can help design non-invasive nerve interfaces while still achieving high application efficiency.
Source : GenK